论文总字数：16035字

摘 要

在这个科技飞速发展的时代，人脸识别成为了人们网络数据安全的一大问题，比如支付宝人脸登陆，各式各样的PS工具层出不穷，外加整容，化妆一些列等等，让人脸数据库变得越发庞大，如果不能很准确的进行识别，会给生活造成很大的困扰，至此，很多研究员走上了对人脸识别研究的这条道路，由于研究人员越来越多，也使得这条路越来越宽泛，在研究人员废寝忘食的研究下，在很大的程度上促进的它的发展，在一般的条件下，基于人的面部性别识别程序大多分为人脸脸部的检测和图片的预处理，面部特征的提取和分类器的设计这三个部分，这篇文章主要针对这三个部分进行研究与探索，还进行了对层出不穷的识别性能比较与分析，大概的工作在接下来一一介绍：

1. 概述人脸性别识别走过的路程，攻克的难关以及目前技术能达到的水准，由于网络的发展，使得获取信息变得便捷，在各式各样的网站上都能找到人脸识别的信息，通过这些大量的信息，特别是最新报道的消息，在进行一定工作量的综述工作，这篇文章重点对目前世界上主流的各种方法进行了概括，归纳和总结，也包括了对人脸识别中的重要部分进行了详细的归纳和总结。
2. 本文主要探索了Gabor变换，它可以用于区域特征提取，Gabor具有两种小波，一个是一维小波，另一个是二维小波，两者具有共同点，同时也有不同点，相同点是两者都具有频率选择性，后者还具有频率方向上的选择性，两者原理会在文章中体现出来。
3. 本文着重在于性别识别，开发工具选择了Opencv，它自带了人脸识别程序，可以利用来实现人脸的识别和提取人脸的区域。
4. 主要成分分析用到了PCA算法，它可以用来对提取到的图片进行维度的降低。
5. 主要解释了支持向量机SVM，它具有一定的分类方法，它依托了结构风险最小化理论，对于男女判别这样的分类问题，支持向量机SVM向学习者们展现了它强大的一面，它透露了很多自己独有的优势与特点，在同等样本的训练和测试下，进行了三种不同的核函数识别准确率的比较，最终确定了以PCA为核心，Gabor为辅助来进行特征的训练和识别。

关键词：人脸检测； Gabor 变换；PCA降维；支持向量机SVM

ABSTRACT

In this era of rapid development of science and technology, face recognition has become a major problem of people's network data security, such as Alipay face landing, various PS tools emerge in endlessly, plus plastic surgery, makeup, and so on. The face database is getting bigger and bigger, and if you can't recognize it accurately, it can cause a lot of trouble in life. So many researchers have embarked on this path of face recognition research, as more and more researchers are working on it. It's also making this path more and more extensive, and in the light of the researchers' forgetfulness, to a large extent, it promotes its development, in general terms, based on human faculties. Most of the recognition programs are divided into three parts: face detection and image preprocessing, facial feature extraction and classifier design. This article mainly focuses on the research and exploration of these three parts. The comparison and analysis of the endless identification performance are also carried out, and the following work will be introduced one by one.

An overview of the path of face gender recognition, the difficulty of the attack and the level of the present technology can be achieved. Because of the development of the network, the access to information becomes convenient and the information of face recognition can be found on all kinds of websites. Through this large amount of information, especially the latest reports, a summary of the amount of work is carried out. Work, this article focuses on the current mainstream of the world's various methods are summarized, summarized and summarized, including the important part of face recognition in a detailed summary and summary.

This paper mainly explores the Gabor transform, which can be used to extract regional features. It has two kinds of wavelets, one is one-dimensional wavelet, the other is two-dimensional wavelet. The similarity is that both have frequency selectivity and the latter have frequency selectivity. The two principles will be reflected in the paper.

This paper focuses on gender recognition. Opencv is chosen as the development tool. It has its own face recognition program, which can be used to realize face recognition and extract face regions.

The main component analysis uses PCA algorithm, which can be used to reduce the dimension of extracted images.

This paper mainly explains the support vector machine (SVM), which has a certain classification method. It relies on the structural risk minimization theory. For the classification of men and women, support vector machine (SVM) SVM shows its strong side to learners. It reveals many unique advantages and characteristics of its own. Under the training and testing of the same samples, it compares the accuracy of three different kernel functions, and finally determines the training and recognition of features with PCA as the core.Keywords: face detection; Gabor transform; SVM; SVM

目录

第一章 绪论 7

1.1 课题研究的目的与意义 7

1.2 人脸性别识别的特征提取研究现状 7

1.2.1 基于Gabor小波对人脸特征提取 7

1.2.2基于主成分分析（PCA）方法 7

1.2.3支持向量机SVM 8

第二章 人脸检测与图像预处理 9

2.1 人脸检测探索 9

2.1.1基于特征方法 9

2.2面部检测实验结果展示 9

第三章 特征提取与降维处理 10

3.1基于Gabor小波的特征值提取 10

3.1.1基于Gabor变换的区域特征提取 10

3.1.2科学探索Gabor小波变换 10

3.2特征提取以及降维 13

第四章 分类器设计 14

4.1引言 14

4.2支持向量机的分类情况 14

4.2.1线性可分的最优分面类 14

4.2.2线性不可分的最优分面类 17

4.3支持向量机SVM的训练算法 19

第五章 实验结果 22

5.1 概述 22

5.2实验平台 24

5.3实验结果与分析 24

5.3.1Gabor特征提取的实验与分析 24

5.4控制台程序展示 25

5.5本章小结 27

第六章 结束语 28

参考文献： 28

致 谢 29

1. 绪论
   1. 课题研究的目的与意义

人脸是人类最重要的生物特征之一，它包含了很多重要的信息。本课题的设计目的是通过机器学习工具实现图像中的人脸定位，人脸特征提取与性别判别。本项目不仅要求具备C 或者Python相关的编程技能，还需要了解OpenCV工具，图像处理和机器学习的基础知识。拟采用的技术路线包括：

（1）一开始，先对图片上的人脸进行检测，简单来说，就是在图片里面找到人的面部，并且把面部区域进行提取，保存作为训练数据。

剩余内容已隐藏，请支付后下载全文，论文总字数：16035字